gh-99891: Fix infinite recursion in the tokenizer when showing warnings (GH-99893)

Automerge-Triggered-By: GH:pablogsal

4 files changed, 24 insertions, 0 deletions

diff --git a/Lib/test/test_source_encoding.py b/Lib/test/test_source_encoding.py
index cfc4b13..b05173a 100644
--- a/Lib/test/test_source_encoding.py
+++ b/Lib/test/test_source_encoding.py
@@ -160,6 +160,18 @@ class MiscSourceEncodingTest(unittest.TestCase):
         finally:
             os.unlink(TESTFN)
 
+    def test_tokenizer_fstring_warning_in_first_line(self):
+        source = "0b1and 2"
+        with open(TESTFN, "w") as fd:
+            fd.write("{}".format(source))
+        try:
+            retcode, stdout, stderr = script_helper.assert_python_ok(TESTFN)
+            self.assertIn(b"SyntaxWarning: invalid binary litera", stderr)
+            self.assertEqual(stderr.count(source.encode()), 1)
+        finally:
+            os.unlink(TESTFN)
+
+
 class AbstractSourceEncodingTest:
 
     def test_default_coding(self):
diff --git a/Misc/NEWS.d/next/Core and Builtins/2022-11-30-11-09-40.gh-issue-99891.9VomwB.rst b/Misc/NEWS.d/next/Core and Builtins/2022-11-30-11-09-40.gh-issue-99891.9VomwB.rst
new file mode 100644
index 0000000..20cd361
--- /dev/null
+++ b/Misc/NEWS.d/next/Core and Builtins/2022-11-30-11-09-40.gh-issue-99891.9VomwB.rst
@@ -0,0 +1,3 @@
+Fix a bug in the tokenizer that could cause infinite recursion when showing
+syntax warnings that happen in the first line of the source. Patch by Pablo
+Galindo
diff --git a/Parser/tokenizer.c b/Parser/tokenizer.c
index ce72e15..463c0e0 100644
--- a/Parser/tokenizer.c
+++ b/Parser/tokenizer.c
@@ -97,6 +97,7 @@ tok_new(void)
     tok->async_def_nl = 0;
     tok->interactive_underflow = IUNDERFLOW_NORMAL;
     tok->str = NULL;
+    tok->report_warnings = 1;
 #ifdef Py_DEBUG
     tok->debug = _Py_GetConfig()->parser_debug;
 #endif
@@ -1201,6 +1202,10 @@ indenterror(struct tok_state *tok)
 static int
 parser_warn(struct tok_state *tok, PyObject *category, const char *format, ...)
 {
+    if (!tok->report_warnings) {
+        return 0;
+    }
+
     PyObject *errmsg;
     va_list vargs;
     va_start(vargs, format);
@@ -2239,6 +2244,9 @@ _PyTokenizer_FindEncodingFilename(int fd, PyObject *filename)
         }
     }
     struct token token;
+    // We don't want to report warnings here because it could cause infinite recursion
+    // if fetching the encoding shows a warning.
+    tok->report_warnings = 0;
     while (tok->lineno < 2 && tok->done == E_OK) {
         _PyTokenizer_Get(tok, &token);
     }
diff --git a/Parser/tokenizer.h b/Parser/tokenizer.h
index 2542d30..16a94d5 100644
--- a/Parser/tokenizer.h
+++ b/Parser/tokenizer.h
@@ -92,6 +92,7 @@ struct tok_state {
                              NEWLINE token after it. */
     /* How to proceed when asked for a new token in interactive mode */
     enum interactive_underflow_t interactive_underflow;
+    int report_warnings;
 #ifdef Py_DEBUG
     int debug;
 #endif